Comments on: Using the ltree module in PostgreSQL http://foobar.lu/wp/2007/06/27/using-the-ltree-module-in-postgresql/ coding should be fun Thu, 10 Mar 2011 18:39:47 +0000 hourly 1 http://wordpress.org/?v=3.1.2 By: exhuma.twn http://foobar.lu/wp/2007/06/27/using-the-ltree-module-in-postgresql/comment-page-1/#comment-3 exhuma.twn Fri, 29 Jun 2007 18:45:44 +0000 http://foobar.lu/wp/2007/06/27/using-the-ltree-module-in-postgresql/#comment-3 What you see here in the postgres queries is probably not how it's stored internally. The simple text-content separating the nodes with dots, is simply a convenience notation. Under the hood it uses some C-code which I did not look at. I don't like reading C ;) But I suspect it's a simple linked list. The reason "ltree" exists is simply because querying a tree using database tables is painfully slow. The ltree module aleviates this. How it's represented internally is not (and should not be) the concern of the database designer. What you see here in the postgres queries is probably not how it’s stored internally. The simple text-content separating the nodes with dots, is simply a convenience notation.

Under the hood it uses some C-code which I did not look at. I don’t like reading C ;)

But I suspect it’s a simple linked list. The reason “ltree” exists is simply because querying a tree using database tables is painfully slow. The ltree module aleviates this.

How it’s represented internally is not (and should not be) the concern of the database designer.

]]> By: GenTr1 http://foobar.lu/wp/2007/06/27/using-the-ltree-module-in-postgresql/comment-page-1/#comment-2 GenTr1 Thu, 28 Jun 2007 13:07:45 +0000 http://foobar.lu/wp/2007/06/27/using-the-ltree-module-in-postgresql/#comment-2 Looks Very interesting... Gonna have a look on that if that can help me to store populations generated during Genetic Programming. One thing I am not to sure is the performance issue. They represent the tree as a whole nested structure. It is actually in spirit quite close to a nested list (python style- assumption: First element of a list or a nested list is always a head node) tree = ['A', ['B', ['D', 'H', 'I'], 'E'], ['C','F', 'G']] I am totally crap with data bases in general, so my question might seem a bit naive, but... Is it not much quicker to store it in the data base as paired elements of a hash map ? (python-style dictionary): tree = { 'A':['B','C'], 'B':['D','E'], 'C':['F','G'], 'D':['H','I'] } Reconstruction of a nested structure would be something done separately, and this way, you can not only store trees but graphs and it should not get too slow to store larger trees. Looks Very interesting…
Gonna have a look on that if that can help me to store populations generated during Genetic Programming.

One thing I am not to sure is the performance issue.

They represent the tree as a whole nested structure. It is actually in spirit quite close to a nested list (python style- assumption: First element of a list or a nested list is always a head node)
tree = ['A', ['B', ['D', 'H', 'I'], ‘E’], ['C','F', 'G']]

I am totally crap with data bases in general, so my question might seem a bit naive, but…
Is it not much quicker to store it in the data base as paired elements of a hash map ?
(python-style dictionary):
tree = { ‘A’:['B','C'], ‘B’:['D','E'], ‘C’:['F','G'], ‘D’:['H','I'] }
Reconstruction of a nested structure would be something done separately, and this way, you can not only store trees but graphs and it should not get too slow to store larger trees.

]]>